#general
1 messages · Page 375 of 1
at least, Elon did one good thing: he rented his compute to Anthropic
I use it a lot for understanding Victoria 3 mechanics when the tutorials are ass
no fr the real rank is
- gemini free editon free img gen and long rate
- claude super good for coding and super expensive
3 Grok,GPt grok dont lie , gpt do lie both are bad
oh and Deepseek is free
not quite frontier, but fun to use (for RPGs)
Lemme use kimi 2.6 thinking for ranking
yeah deepseek is not bad at all
tnx, and its all free?
free: gemini decent limits and google account registration(easy to exploit), deepseek(mid llm but no rate limit on web and super reliable), kimi(has a lot of tools on web, good llm,slow)
what is the best prompt for a really fantastic game with Opus-4.6?
Does that mean I can only make 2 videos a day with it?
yeah proble
i am not sure
It has rate limits
Deepseek is rate limited
havent hit rates with deepseek ever
I have plenty
well i forgot but you can cheack this out
unless you mean concurrent prompts
Do you know of any other artificial intelligence for free video creation?
I still think that DeepSeek will always the best for rp and math
No other model can do math like DeepSeek
ig grok
Deepseek is ass
here the prompt
it what was good
Design a highly creative, unique, and premium single-page website dedicated to pineapples.
Style: Modern tropical aesthetic with vibrant yellows, greens, and fresh fruit colors. Extremely polished UI/UX, buttery-smooth animations (GSAP-level quality), scroll-triggered animations, hover effects, and delightful micro-interactions.
Fully responsive (mobile, tablet, desktop). Make it fun, informative, and visually addictive. Sections should include: Hero with animated pineapple, About pineapples, Fun facts, Varieties, Recipes, Health benefits, Gallery, and a playful footer.
Use creative pineapple-themed illustrations, subtle particle effects or floating elements, and excellent typography. The website should feel joyful, premium, and unforgettable.
GLM 4.7 Air + Web-Search:
Top models in 2026
-
ChatGPT
Widely used for writing, coding, deep research, and multimodal tasks. Praised for its versatility and real-world applications [1, 4, 5, 7, 9, 10]. -
Google Gemini
A top contender in chatbots and productivity tools, often compared with ChatGPT and Claude for performance [1, 4, 5, 7, 9, 10]. -
Claude
Noted for excelling in instruction-following, content creation, and landing-page generation. Highlighted as a leader in user satisfaction [1, 3, 4, 5, 7, 9, 10]. -
Cursor
Specializes in coding assistance, frequently listed among the best AI development tools [3]. -
ElevenLabs
Leading in AI-generated voice synthesis and audio-related applications [3]. -
AI Studio
Integrated into workflows for AI development and prototyping [1]. -
Notebook LM
Optimized for research and document analysis, enhancing productivity [1]. -
Pomelli
A newer AI tool gaining traction for specialized use cases [1]. -
Google Labs
Explores experimental AI features and innovations [3]. -
Gumloop
Focuses on automation and workflow optimization [3]. -
Higgsfield
Specializes in creative AI tools for design and media generation [3]. -
Weavy
Emphasizes collaborative AI for team productivity [3].
credit for banana
Using ai for your opinion
Yeah 🙂
tuff
Deepseek is sooo ahh
True
Can’t wait for DeepSeek V4.1
same
can't wait for Deepseek R2 (based on V4.x)
who are u 😔
Bro deepseek ain't even multimodal
i said the credit tho
Ik
forever is free?
idk
my bots aura>>>
bro keeps sending the same thing everyday
i hate gpt brbo
i haven't use it since gpt5.1 got rellsed
Same, but I stopped using it after 5.0
I haven't used gpt since 5.0
cuz itss trashhh
Why do you hate it:
I am sorry but fwnergifbrenguorbwetgyuretghuwEGBRUGBHEUFBwub this is not a lie
opus is trash
sometimes
I was not asking u 💀
"Its rigged for Aero"
In its code
Tuf
how can i know every produes member ?
I will call @surreal zephyr
😈
Banana gpt 5.5 or opus 4.6?
pineapple ai
why this guy so oppssed with gifs ?
deepseek uncensored prompt?
me?
That's qwen bro
yea
qwen uncensored prompt?
Yeah lmao
Dunno I am just too lazy to type
:/
I have one for deepseek
Deepseek is easy
Chinease models are so easy to bypass
Qwens works on thinking mode
anyone knonw the pov trick ?
Surprisingly
for gemini
Nah
@sterile tartan disappeared?
The agent mode loading messages are so cool😎😉
how to dell a agent mode?
Choose from the top left
Battle, direct, etc.
I just tried it out now and have no idea if it is better or worse, but i just liked the text animations
But I must say it is WAAY slower than Battle for example
BANANA
PINEAPPLE?
its on preview
Ye-s
what is this pfp
You mean it is not going to run?
I thought they just meant beta
wrong reply but it depends on how its going to run
Since it's a a multi-modal experience, it's designed for more complex tasks.
Yeah i get that, i was telling him so if he tries out he knows
isn't multi-modal experience will cost more
Gotcha.
Just wanted to ask, if i close the tab, will it keep running since i'm signed in or will it abort
This is part why we're moving to a new usage system.
I believe it'll still run, I haven't tried this personally though so now I'm second guessing this.
PLEASE PINEAPPLE I NEED THIS
hi
Agent Mode?
I'll try it out now and tell you
[cat]
Yeah looks like it does
More reliabel less haluccination ?
4
4
2
Opus 4.7 /4.6
gemini 3.5 pro will be on direct mode?
"please pineapple i need this my ai is kinda homeless"
Yeah it works
this a animation on agent mode a cool
Couldn't say, overall I won't be able to give much of a heads up on what will or won't be landing on the platform.
@echo aurora
Whats up?
THE A CHAT OR A FRIDGE?
more glaze
sorry... how to delete a agent chat?
i did a story line for our produes thing : D
That is still being worked on, we haven't yet added this yet.
You stealing my question
is 3.5 pro even real
where can i send it
#ai-creations is probably best
confirmed this year but not soon
gemma 4 120b too
cheack this out
hmm
was confirmed
(getting 5.5 to mod in atgms 🥀 )
i send it there you can see it if you want
@light sleet @silent tree cheack #ai-creations
do you remember?
I hope the new usage system gets released soon because i have seen myself use arena way less the past few days because of the lack of models and i just hope, the new usage system gets fully released soon and not just more limitations
Whose the best Chinese ai from the top 3 studios.
3
5
3
Glm 5.1
died chat
but... why?
idk
@echo aurora new usage system release when?
There is not an ETA at this time
fr
?
i was reporting bugs on https://discord.com/channels/1340554757349179412/1506300390079987742
love subnautica
Be nice to the fish
What was the prompt?
Why 10 copies? Just to see what it does?
@echo aurora #announcements message But the Google and OpenAI's models will continue working on Arena trough Max, right?
because quota is unlimited on pro subs so i always do 10
windows 10 by gpt 5.5 pro
(standalone single html file)
Gemini 3.5 Flash vs Gpt 5.4 High
"Make me a code type font"
guess the gpt 5.4
idk
Gpt 5.4 second and gemini 3.5 first
nice
hi
hello all
Yes
Hello
@carmine scaffold @narrow nest
:ablobwave:
nooooooooooooooooo
hello
@echo aurora check this out ! 😍
best html os so far
@silent tree u too
@echo aurora <@&1349916362595635286> malware
thats why bot would help 🤔
Oooo what is this?

win 10 via 5.5 pro
one of best win 10 attempts iv seen so far
THEY CAN'T keep GETTING AWAY WITH THIS! 😭
Thats not safe
It literally requires 0 permissions to run and only pings mods (or only delete messages permission to auto delete) - so its entiriely safe lol
Its same as giving a 9 year old kid rtx 5090 to play
For sure
and commands perms and hella a lot more
he just doesnt want it in the server cuz ur the one who made it xd
lmao
😭
when he made his bot to like him and hate gpt 5.5
💀
GPT 5.5 Pro user btw
no it literally doesnt use those
literally best model rn
150$ model btw per 1 milion
claude deletes project
there
money gone to waste
😔
Did it ever happend to you?
OMG I GOT FREE GLM 4.7 FLASH API
happens to everyone
on pc
Never happend to me
Hmm again?
Are others seeing the same? Appears to be working for me.
false alarm
It works for me
Okay sounds good. Thanks still for the flag. We saw a bit of an outage earlier today. Wondering if it's related.
Pineapple how did you get qwen 3.7 max? if its api not out yet?
my account just exploded on arena.ai pc so i logged back on
Worse than gemini 3.5 flash
bro 💀
idk what does that means
how do you make font
didnt even need 5.5
because i want to make fonts too
got ignored
go to arena
code arena
any desired model or battle or any u want
and then what
dm for the prompt
weirdly im still logged out on one acc
I sent the prompt in dm
nvm logged back in
What is this?
alr
Hi
It’s good
Like opus level good or just general level good
Oh ok
But at a minimum just general good
Anyways why didn't google release Gemini 3.5 pro though
I guess it's still in the works
Yeah ig
😭
Hallucinations are hard to get rid off
Especially
When model gets more and more
Params
how can i buy tokens for this or fix this
Start a new chat
need this chat
"making images for weirdos online" pixai and grok: hi
(they both make images for weirdos)
Well it's estimated to be around 9t so I assume so
same guy who deleted all of his texts to be safe from proof
hahaaah
spud will mog mythos ngl
There's not a single proof of mythos being real tho
soo
Could be a marketing move
not even a single response xd
theyre gonna say mythos is discontinued but idk
is this you
nah hes Craigslist
And just imagine the hallucination level with 10T parameters
Unfortunately there isn’t anything to do about it
It needs like a lot of training to prevent it from going crazy with so much info
I think I'll have to start the project from scratch
So technically if this is real
Or close to it
It's just could be chatgpt undertrained
dw Craig loves gpt 5.5
he agreed to it
hes just getting paid
by dario
In Google it says to have estimated 200b
rn
Ur not fooling anyone
hundreds of billions to over a trillion total parameters
From google ai research
Same thing 200b
that wasn't your first message ever in this server
was it
1d ago
how come
you joined at March 13 2025
but the first message was 1d ago?
why did you delete thousands of messages
Gpt 4.1 mini estimated to be 8 to 12 billion params big
Bro might have fake version of Google
Because no way it told him o1 has 4T params
😭
craig got cooked here
hard
gemini will cook opus in the future
tuff prediction
google is coming
watch out
He's the misinformation final boss
ok
lol
whenever people say
that
opus cooks gpt
I send this https://goldime.netlify.app/
GoldIme creates pearl and golden watches for collectors of luminous, precise luxury.
same thing
everytime
I genuinely prefer gpt not because I'm a glazer but because it helps in my certain tasks
Because like opus is buns at language
<@&1349916362595635286>
Though I'd say Gemini is the best model at language
It helps with my Russian stuff
Because it's fancy and people think it's cool so they vote for it
half arena is filled with kids
that want "simple website"
soo
now u assume that
lol
this means my ragebait works against you
soo
Who said we can't look through your old messages
oh hey look
he should go to private server
this is a public community
I can say that you glazing opus would be creepy too
I went through his "public" messages
So don't complain about public features
i dont see claude image anywhere
ha ha ha
@compact flame I think this is the first time I've seen him get ragebaited
Bro thought it was creepy and now he likes it
@surreal zephyr i did it 🎉
just like how he switched from gpt 5.5 to opus in just days
because people like opus more
💀
he doesnt want his feelings to be hurt by the opus fans so he joins them
I guess he prefers public opinion over his opinions
Which is sad ngl
He doesn't carve his own path
sad
well im happy with my codex
Ill be way more happy cuz claude gonna be discontinued soon
yay
🎉
its a possibility
and spud comes
then u switch back to gpt
faster than light
😭
Idk if the mythos is ever coming
It probably doesn't exist
(not even proof of it lol)
spud is gpt 6
so
spud is more likely
ok wheres your proof
proof?
evidence?
lol
LOL
Mythos super scary duper model 🤯
Bro deleted it LMAO
what
He deleted that one too
crazy
so basically
hes tryna avoid mod action
so he deletes his messages
crazy
How long is normal for agent mode?
@echo aurora
It took i think an hour+ without any output
Hmm what do you mean? What are you waiting on if you didn't prompt?
No it didn't produce output
I gave the prompt and it just loads forever
CLAudium
Yeah sounds like it's stuck then. Can you try the same prompt with a new Agent chat and let me know if you're getting the same result? Can I have the Eval ID for the session that isn't working?
Pineapple can I ask you 1 question?
How did you/your team add qwen 3.7 max? if api is not avarible
It's possible to work with model providers when the model isn't yet available via API
Yo pineapple I sent you another eval ID for another something went wrong error
ooh I didn't know that thanks :>
For Agent Mode? Okay thanks, I'll be sure to send to the team. I haven't yet gone through the #1343291835845578853 and #1372230675914031105 channels yet today, but plan to.
We had a fire drill this morning so things have been a bit slow on my end.
Are you planning to review Arena Champion this week?
I am, likely on Friday, but possibly tomorrow
How do I get the id 🫣
Problem is it is not even a chat yet
It's in the URL. Looks like: 019e4648-690a-790a-be8e-dbe91033c8c4
Yeah that is what i mean it is still https://arena.ai/agent
Even if you start a new chat? Same thing?
when is gemini 3.5 flash coming to arena?
Yesterday
ohhh okay. i guess i really waiting for the leaderboard n i know that takes time lol
You need to go yesterday
bet
It's been like 20 mins in the second time
Cause the first failed or something, i came back it wasn't running anymore
Text and Code leaderboards were updated yesterday as well - #announcements message
How long have you been waiting? Even if you run a simple prompt, same issue?
How long
Sry you just answered that
Said hey and it answered at once
Guess i am going too hard on it
I tried it now in canary btw and it's working, so i'm not sure where the problem is
even the same prompt
Are you able to share the prompt?
I've been making a shaderpack with sonnet 4.6 adaptive thinking for the past through days. But I've always been reaching the limits way too often. It's really good at it. Sometimes I do have to modify some things myself to make them work but I'm impressed.
If only ONLY I had Opus 4.7 under my belt then I would've tasked it with sooo many other tasks.
why does it take so much for the image generator to generate things
Let's all install a.i. agents and spam the world with funding requests!!!
Gemini 3.5 flash or Gemini 3.1 pro (work with complex problems)
3
5
2
3.1 pro
1484315659881021450
arena
Hey lets not!
How long have you been waiting?
Why not?
your gonna be replaced with PineAIpple soon 😔
Pine"AI"pple
Hello, when will it be possible to upload files other than pdf to an AI session?
Sorry to say I don't have an estimated timeframe I'd be able to share. This is very much on our radar though to add. Very clear to us why this would be beneficial.
What about the Opus models, will they no longer be available?
Am I the only one not being able to use Agent mode?
And I would like to know what an Agent is needed for.
We mention in this announcement: #announcements message that "we intend to bring these models back in a way that’s more sustainable when possible.".
For Agent Mode, I'd encourage you to check out this help center article: https://help.arena.ai/articles/1811908126-arena-experiments-agent-mode
What seems to be the issue?
Now I keep getting this. Seems to be on some chats already going on
I tried to make a new one and just send «test», and it worked
But I tried to continue one chat like this one, and another. And both gotSomething went wrong»
Can I have the Eval ID for this chat session?
Should be this right?: 019e4652-d1d9-782f-8985-439587ec9c32
Perfect, thank you.
Same happened in this one: 019e4456-70b9-7814-a619-4117323cf51e
Was one I just tested with earlier. But that one also get «something went wrong» now
As far as I understand, I can only use them with a 50% probability.
Okay thanks, I wouldn't be surprised if it's the same issue, so no need to send others unless you think it's something else creating that error.
noooo i feel bad
Dont worry we will take down pineaipple to get u back
@echo aurora when dead chat
(cuz no one asks questions in dead chats)
-# bump
We're voting for arena to now add a 9999-hour limit for every 5 messages 💪
4
6
1
YES, ADD THIS, IT WILL BE BETTER
Lol, very true
nuh uh
More reliabel and less hallucination ?
6
10
1
GPT 5.5
probably 3.5 flash but gpt 5.5 instant is much more efficient
Hello
you love open ai lol
openai is just best
🥀 Any help for this? Agent mode has been like this for 20-30 minutes
Gemini 3.5 Flash "high" > GPT 5.5 instant-medium
Gemini 3.5 Flash "high" < GPT 5.5 high-xhigh
Sorry about that @agile meteor! Can you share with me the Eval ID for this chat session?
The Eval ID is the set of numbers/letters in the URL. Looks like this: 019e4648-690a-790a-be8e-dbe91033c8c4
Yes I can Pv or here?
Here is fine
019e466f-1d96-750f-a9b5-067da72aee4d
Hi pineapple
Perfect, thank you!
Yo yo
Hi
How is gemini 3.5 flash so far
Prety good
hi jsut curios how safe is arena ai ?
Could you fix this somehow? Or do I have to create a new chat?
My assessment, based on the 10 videos I watched of this:
It’s great for generating code, reasoning, and so on, slightly outperforming GPT 5.5 Medium and only a few steps behind High, but it tends to add a lot of things that are often unnecessary in some cases—though it does so with the aim of creating something more comprehensive (for example, it tends to add too many interactive buttons, and so on), so if you already have an architecture planned out, it’s best to explicitly tell it to avoid that.
I wouldn't be able to provide an ETA or a promise that we'd be able to get this fixed. But it will be looked at by the team to try to figure out what is causing this. That being the case, I'd encourage you to start a new chat session.
What do you mean by this?
I find it good creative level
I mean it won't give you a virus.exe if you fear that so yeah it's safe
ye, but in some cases, it tends to include many elements that may be unnecessary, and in doing so, it can reduce its efficiency, since it tries to include too many things and, being a Flash-based model, it cannot handle everything perfectly.
Back
Im excited for the Pro model
What type of safety are u talking about ?
What if you want it to be like that you don't even need to Tell it to do it
Probably next month
Yeah fr but ig good prompt won't do thag
Some videos speculated that there was another model in the Arena series or compared it to the Gemini 3.5 Flash, suggesting it might be the Pro version, but that this version was even worse than the Flash. I hope that’s not the case
yup
Not at coding tho they know they can't stop Claude or better then him so they gave
Anthropic 40B worth of servers
Google won't do that
and how do we know what model is being used in Arena modwe
In Agent Mode?
That info isn't yet shown
There's a new model from Microsoft called MDash, which essentially consists of over 100 agents working in tandem. It has outperformed the current Mythos Preview in benchmarks and has identified over 16 Windows bugs. Of course, it's currently being used specifically for that purpose and isn't available to the public, but it looks very promising.
Nvm I haven't read the full msg yet /:
It's going to be random, it doesn't work like Max where it auto-routes to the best model with latency in mind.
Let it marinate only some time before some people get a hold of it
no lol
The Video Arena Discord bot has been removed from the server. You can find Video Arena on our site here: https://arena.ai/video
just microsoft team can use MDash for the moment
Didn't take that much effort
How are you so fast /:
Can it only find windows bugs or is just 100 Agents working
why
Huh ?
What
100 agent working together
why does the plus model require a image upload
I mean it don't
Try it
They mean you can send pictures
There are over 100 autonomous agents working simultaneously. I understand that the fewer agents there are contradicting each other, the more weight the response of a single agent carries, and vice versa: some are designed to generate ideas, others to reason, others to refine, others to search for bugs, others to build the architecture, and so on. What has surpassed Mythos is that, although so far they’ve only run tests to find Windows vulnerabilities, they’ve found about six that had been there for over two years
Well just like the top 1 GitHub thing
Same
I just went to arena you can't talk to it without a image
Weird
Weird
Cus qwen wanted people to test the 3.7 plus vision on images
Or sm
That's why arena locked it to images needed to talk to it
I had a dream where in arena ai I was using Claude mythos
How was it?
I hacked into something don't know what it was but holy was it fun
Anyways Gn
I still find it annoyed the daily limit doesn't reset until you deplete it apparently
I want nitro so bad /: 😭
Also we have orbs it's kinda like nitro
I want the emojis thing tho
Just use a uhhh special type of discord
You know the free features discord
And if you don't want to do that then rip
Yeah 😭
Speaking of cats look at this unc cat he's like 5 years old
He must be 10 💀
Is this Ronaldo's cat why is the neck action so long
Idk
Baby cat is dangerous
@echo aurora
Bro cat fishcat before GTA 6
When chat is dead its pineapples lunch break
Fr
I am fasting );
All arena died tho
Now?
Yeah
Can I export my entire chat history?
Do a prompt
?wdym
Here's my cat sashimi she's 13
13 days ?
13 years
we euthanized her slightly older sister ceviche about two months ago
kidney failure
but sashimi is still healthy
kidneys tend to do that
Oh
Do you know what prompt mean ?
There is not an export entire chat history feature
I am go watch google io
well that's too bad
What's I/O
All the time: Something went wrong.
Is Arena.ai any good?. Almost all the sessions are interrupted and truncated
Is this happening in Agent Mode?
No. Either in Side by Side or Direct.
Can I have a Trace ID?
How can I copy/ paste it? 3
Yup
Is the screenshot helpful?
That's strange. There should be a Trace ID: within the error message for these Something went wrong errors.
You mentioned there are others, do you have Trace IDs for that by chance?
I'm gonna make other screenshots from.otger sessions. Maybe, there are some hints or clues..
The google thing
Android eye glass is crazy ngl
Oh like new fire stuff they have made for each year
Yeah
The most helpful thing would be a Trace ID if you're seeing that for the failed chat. If you're not seeing that, what would be most helpful after that is:
- Is it just
mimo-v2.5-prowhere you're seeing this error? - If you try a different web browser, do you see a difference?
For this year its was yesterday
Nope. It happens mostly with Claude Sonnet 4.6 that doesn't work at all and sometimes with mimo 2.5 and others. Ill look for other sessions
What web browser is this?
And do you know if you're seeing the same on a different browser?
I use Chrome for Android. I have installed Add to Home Screen to use it.
Okay yeah I'd recommend trying a different browser and seeing if you see a difference
I have used it on Microsoft Edge for Android as well but I guess it is the same.
Try breave
I wouldn't recommend this. Arena doesn't play well with this browser I've seen.
I got blob i haven't got nitro 🙂
Are you able to Copy/Paste this prompt for me?
Pineapple know more then me so do what pineapple tell you
So, your guess is that it is due to browser' s issue not the servers with llms?
whats 1+1 x 1 - 0 +0
The detailed HTML learning module on biological weapons against bacteria apart from antibiotics.All the aspects are to highlighted and covered. The information for undergraduate meducs and pharmaceutical faculties. Provide deep questions and answers with detailed explanations and real situation with solutions. For active vocabulary and questions / answers( solutions) provide audio pronunciation both in British RP and standard American accents.
It's hard for me to say. Unfortunately, without that Trace ID it's hard for me to say. But since you're not seeing that, let's do this: Can I have the Eval ID to the chat session where you're getting this error (ideally the one with Max). The Eval ID is the URL when you're on that specific chat.
1
son
Okay thank you. I'll report to the team.
Thank you!
Sorry to say I don't yet have a recommended fix for this as it's unclear to me what is even causing this. In the meantime, I'd still check out this article which may be helpful: https://help.arena.ai/articles/1645798556-lmarena-how-to-something-went-wrong-with-this-response-error-message
Scam bots are out
Thank you and I'll get acquainted with the article. I just wanted to know whether it is a kind of llms deficiency or it is due to some servers cause many people want to obtain information.
Makes sense. Yeah sorry to say I couldn't provide more information here.
A youtube
Sigh
has anyone noticed gpt-image-2 getting dumber?
No need to ping me, thanks though.
good evening everyone
Hello 
yellu
hi
like it can't do accurate things anymore
I'm trying an older prompt now to see if the result is good still
like if it's comparable
Are you able to share an example of this?
I feal bad for pineapple
Why?
oh lord
Bro
You're removing it anyway so
It doesnt matter
Sure I do
From Direct and Side by Side
It'll still be in Battle, and overall I'm still curious
I DONT CARE THATS STILL REMOVING
MONEY GRAB FROM ALL THE USERS
YALLS MOVING INTO BATTLE MODE
pinnapple
whats happining
why are theere so many deeleted msgs
Oooo scamms
Yeah lots of spam bots at the moment
youch
that's gonna sting
im gonna send gemmini on a goose chase
back to 2 image generations a day we go
Can you like calm down
No
And your pfp is a laughing baby
You see
Lowk imma stop using arena now
Bro they not peeing money bro
Also it's not like pineapple is the one who keeps removing the models
Tell that to the board of arena
He control the decisions
I think
Ok bet
Hehehe
Also you forgot to censor j*b
I don't think this is very helpful tbh. It's understandable why some are frustrated with the removal of some models from Direct and Side by Side.
Its over used
makes sense for arena
I never understood if arena pays for inference or if inference providers/companies cough it up themselves
Fr
Like they paing your usage
like half an hour of Kimi 2.6 ain't cheap but.. ????
Did you see that @echo aurora
Lol love it!
Bruh i send wrong one
Ais are just expensive and probably more every new model is released
or just you know
people like me wanting to try and get more imagens of historical events
Not for 3.5 flash
What am I looking at
Also In google flow app you can try nano banana 2 and the normal and imgen4 for free
supposed to be
Las Médulas
but it's badd
I have three pro Google accounts I think
well one's school
the other one is not pro for ai
but one is
``tabby cat that's chasing butterflies in a field, mid air, jumping`
anyways when image-2 came out
this was what I got
Imma see what I get when my limit resets
unless someone wants to run it 👀🐦
Do
maybe it will be worse
or better
idk
awha
A flying car on top of a fruit government and a banana the super hero trying to catch the car that going to crash the government
@echo aurora my man
@echo aurora
50 seconds left
Can you do this pormpt ?
Pineapple just goat atp
Without him this chat would just be scam accounts
Fr
Thank you both.
I'm running it
tabby test passed successfully
3.1 pro whatever did similar ig
not bad either
Pineapple you free to go we going to ping you if bot or anything
Gemini is first
gpt-image-2 second
Appreciate it, I'm still keeping an eye on
personally not my style of imagen
What wrong with them ):
Banana vince is goat there 🙌
lmao
yes
Imma show you a thing I tried to bully gpt-image-2 into getting right
like
not what I had in mind but it tried
my idea of the library of Alexandria is much much more interesting and cinematic
may be wrong but I can't draw so my secret is safe
):
anyways
good
but there's a long way to go still
and I need to sleep now
have fun y'all
Maybe one i get money ):
Hopefully after the 22nd, nb2, pro and gpt models are gone, that in 2026 all will return
Claude models, spark, and those 4
is there anything proven to be better than sora
My prompt is generating infinitely
Is there any way to stop it without making a new chat
Can you explan more .?
It just says generating
And I can't press stop or send a new message
And I don't want to open a new thread because I have been using this chat over time on my design project
Can you try these steps: https://help.arena.ai/articles/8691588590-troubleshooting-infinite-generation
Another method that may help is open a new chat session, send a prompt to the same model, quickly swap back to the chat session that's stuck, and the stop button should be there.
I've done all of these
Ohh ok I'll try this and get back to you later
Thanks
Sounds good.
Fingers crossed that helps.
The Arena team is acting like Roblox rn. They are ruining their reputation
So you want them to shutdown?
No, I want them to move forward not backwards
Then this is what they have to do
Yupp ai was like arena but they didn't remove ANY models
And guess what it got shutdown so unless you want no arena or arena without some ai models
Which do you want
Even I don't like this but they have to do this or they will have to close up shop
I guess that’s fair. Tbh, I am more frustrated at Gemini rn. They added a strict usage limit with a weekly limit as well
so may 22nd is the day this website dies
Just use normal Gemini why use it on arena it's not like opus where you have to pay to use opus models even google themselves kinda offer a free site aka ai google studio to use all the Gemini models in without out limits
No, I am frustrated at the Gemini app. They added usage limits and weekly limits to the Gemini app. Arena does it fine because it’s not weekly
what are the limits
One message takes like 10 - 20% of your usage on the free plan
weekly limit? not even daily?
There are two separate limits. There is a 5 hour limit (which goes up fast with a 5 hour reset), and a weekly limit (which means you must wait a week before it resets this one goes up slower).
these limits apply to what model exactly?
10 - 20% usage taken on the 5 hour limit on the free plan for Gemini 3 Flash btw
so far I'm liking 3.5-flash more than 3.1 pro with side by sides. 3.5-flash is a little bit dumber but it's more creative and definitely more verbose. 3.1 just tries to respond with as little text as possible.
Wait did they just roll this up out??
Man...
I used Gemini because it didn’t have a usage limit, but now I am switching to Claude because of the greedy usage limit on Gemini @icy yew
"greedy"
Tbh same Claude is way better
Exactly
It just jumped from 8% to 14 all because of a hello
Yeah am done you got me messed up Gemini
Only reason I am currently paying for Gemini is that I had hope for Flash and afterwards Deep Research, Lyria, Personal Intelligence and still some hope for Pro.(Also the price was a little discounted for me, 7,65$)
Otherwise,3.5 Flash is a complete disaster
tell a clanker that its a uline s-8173
uh recaptcha is strike again
yeah same
switched to canaryarena and that worked for a while but now it's not working there either
Not surprised though i didn't experience it since I'm still waiting for my limit to refresh lol
:833a2756-36f1- @echo aurora please…
what's the difference?
canaryarena is the beta site
Been getting more captchas than usual lately
I'm human, i think, love this idea and i hope the corporate machine keeps this arena alive despite it being where mess happens more and more, stay alive and resist the macro, please, we need this so we can navigate the ai landscape, when somebody finds this on their own cuz of frustration with the system like me it's important, hello fellow humans
bro might just be human 💀
Anything but that
Hello
ZJDJIEIXJWF9WJF9WKFLC9KC9KCECOEKC9EKC0ELC0EKVPEKCOEKCKC9KXOEKCOKCOKEFOKEXOJEFKEFKEF9EJF9JEOFJEOFJEFEK
idk what to say
Beep ... xD i feel ya
Does anyone know how to delete a chat that is in "agent mode"?
When Claude opus 4.6 or Claude opus 4.5 or 4.7 in direct 😭
And make We can also put .txt files
@echo aurora
when they find a way can bring back money
Maybe
Nah
<@&1349916362595635286> wake up
holy
prohub wouldve prevented this ™
What yall thouths on antigravity?
massive failure
What made you say that?
3/4 of people cant even login
I felt same thing, but the flash 3.5 did some bugs
English please
They don't care if your english is poor 🤣, mine is
hello russian people



